FORUM: QUALITATIVE Volume 6, No. 2, Art.

27
SOCIAL RESEARCH May 2005
SOZIALFORSCHUNG

"Emergence" vs. "Forcing" of Empirical Data?

A Crucial Problem of "Grounded Theory" Reconsidered

Udo Kelle

Key words: Abstract: Since the late 1960s Barney GLASER and Anselm STRAUSS, developers of the method-
grounded theory, ology of "Grounded Theory" have made several attempts to explicate, clarify and reconceptualise
induction, abduc- some of the basic tenets of their methodological approach. Diverging concepts and understandings of
tion, theoretical Grounded Theory have arisen from these attempts which have led to a split between its founders.
sensitivity, coding
paradigm, theory Much of the explication and reworking of Grounded Theory surrounds the relation between data
building and theory and the role of previous theoretical assumptions. The book which initially established
the popularity of GLASER's and STRAUSS' methodological ideas, "The Discovery of Grounded
Theory", contains two conflicting understandings of the relation between data and theory—the con-
cept of "emergence" on the one hand and the concept of "theoretical sensitivity" on the other hand.
Much of the later developments of Grounded Theory can be seen as attempts to reconcile these
prima facie diverging concepts. Thereby GLASER recommends to draw on a variety of "coding
families" while STRAUSS proposes the use of a general theory of action to build an axis for an
emerging theory.

This paper first summarises the most important developments within "Grounded Theory" concern-
ing the understanding of the relation between empirical data and theoretical statements. Thereby
special emphasis will be laid on differences between GLASER's and STRAUSS' concepts and on
GLASER's current critique that the concepts of "coding paradigm" and "axial coding" described by
STRAUSS and Juliet CORBIN lead to the "forcing" of data. It will be argued that GLASER's critique
points out some existing weaknesses of STRAUSS' concepts but vastly exaggerates the risks of the
STRAUSSian approach. A main argument of this paper is that basic problems of empirically
grounded theory construction can be treated much more effectively if one draws on certain results
of contemporary philosophical and epistemological discussions and on widely accepted concepts
developed in such debates. This especially refers to the critique of naive empiricism, to the concept
of hypothetical or abductive inference, to the concept of empirical content or falsifiability of
statements and to the concept of corroboration.

© 2005 FQS http://www.qualitative-research.net/fqs/

Forum Qualitative Sozialforschung / Forum: Qualitative Social Research (ISSN 1438-5627)
FQS 6(2), Art. 27, Udo Kelle: "Emergence" vs. "Forcing" of Empirical Data?
A Crucial Problem of "Grounded Theory" Reconsidered

Table of Contents

1. How Do Categories "Emerge" From the Data? "Theoryladenness" of Observations as a Problem

for Grounded Theory Methodology
2. Different Approaches in Grounded Theory to Solve the Problem
2.1 GLASER's approach: theoretical coding with the help of "coding families"
2.2 STRAUSS' and CORBIN's approach: axial coding and the coding paradigm
3. The Split Between GLASER and STRAUSS in the 1990s
4. Towards a Clearer Understanding of the "Grounding" of Categories and Theories
4.1 Abductive inference as a logical foundation of theory building
4.2 Empirical content or falsifiability as a criterion for the applicability of theoretical
preconceptions in qualitative inquiry
4.3 The necessity of corroboration of empirically grounded categories and hypotheses
5. Conclusive Remarks
References
Author
Citation

1. How Do Categories "Emerge" From the Data? "Theoryladenness"

of Observations as a Problem for Grounded Theory Methodology

Can the claim to discover theoretical categories and propositions from empirical
data be reconciled with the fact that researchers always have to draw on already
existing theoretical concepts when analysing their data? In the past three
decades following the publication of GLASER's and STRAUSS' famous
methodological monograph "The Discovery of Grounded Theory" both authors
have made several attempts to make these two conflicting methodological
requirements compatible. [1]

One of the main purposes of GLASER's and STRAUSS' "Discovery book" was to
challenge the hypothetico-deductive approach which demands the development
of precise and clear cut theories or hypotheses before the data collection takes
place. GLASER and STRAUSS criticised the "overemphasis in current sociology
on the verification of theory, and a resultant de-emphasis on the prior step of
discovering what concepts and hypotheses are relevant for the area that one
wishes to research" (GLASER & STRAUSS 1967, pp.1f.) and bemoaned "that
many of our teachers converted departments of sociology into mere repositories
of 'great-man' theories" (ibid., p.10) leading to an antagonism between
"theoretical capitalists" and a mass of "proletariat testers" (p.11). Thus the
Discovery book was an attempt to strengthen the cause of researchers and
doctoral students who formed this scientific proletariat:

"(...) we are also trying, through this book, to strengthen the mandate for generating
theory, to help provide a defense against doctrinaire approaches to verification (...). It

© 2005 FQS http://www.qualitative-research.net/fqs/

FQS 6(2), Art. 27, Udo Kelle: "Emergence" vs. "Forcing" of Empirical Data?
A Crucial Problem of "Grounded Theory" Reconsidered

should also help students to defend themselves against verifiers who would teach
them to deny the validity of their own scientific intelligence" (p.7). [2]

GLASER and STRAUSS proposed a "general method of comparative analysis"

which would allow for the "emergence" of categories from the data as an
alternative to the hypothetico-deductive approach in social research. "We suggest
as the best approach an initial, systematic discovery of the theory from the data
of social research. Then one can be relatively sure that the theory will fit and
work" (p.3). Following the Discovery book a crucial measure against the forcing of
data into a procrustean bed would be to "literally to ignore the literature of theory
and fact on the area under study, in order to assure that the emergence of
categories will not be contaminated ..." (p.37). [3]

Ironically, such a stance represents one of the roots of positivist epistemology. In

the early days of modern natural sciences in the 17th and 18th century the most
early empiricist philosophers like Francis BACON or John LOCKE were
convinced that the only legitimate theories were those which could be inductively
derived by simple generalisation from observable data. Following BACON, one of
the most important tasks of an empirical researcher was to free his or her mind
from any theoretical preconceptions and "idols" before approaching empirical
data. However, since Immanuel KANT's sophisticated critique of the pitfalls of
early empiricism (nowadays often called "naïve empiricism" or "naïve
inductivism", cf. CHALMERS 1999) this epistemological position has lost most of
its proponents—and even most of the followers of "Logical Positivism" in the
1930s did not adhere to it. The idea that researchers could approach reality "as it
is" if they are prepared to free the mind from any preconceived ideas whatsoever
has fallen into deserved bad reputation in contemporary epistemology.

"Both historical examples and recent philosophical analysis have made it clear that
the world is always perceived through the 'lenses' of some conceptual network or
other and that such networks and the languages in which they are embedded may,
for all we know, provide an ineliminable 'tint' to what we perceive" (LAUDAN 1977,
p.15). [4]

It is impossible to free empirical observation from all theoretical influence since

already "(...) seeing is a 'theory-laden' undertaking. Observation of x is shaped by
prior knowledge of x" (HANSON 1965, p.19). Since the 1960s it is one of the
most crucial and widely accepted insights of epistemology and cognitive
psychology that "there are and can be no sensations unimpregnated by
expectations" (LAKATOS 1978, p.15) and that the construction of any theory,
whether empirically grounded or not, cannot start ab ovo, but has to draw on
already existing stocks of knowledge. At the same time this philosophical critique
of inductivism and the emphasis on the "theoryladenness" of observation also
highlights the role of previous knowledge in hermeneutic Verstehen (KELLE
1995, p.38): Qualitative researchers who investigate a different form of social life
always bring with them their own lenses and conceptual networks. They cannot
drop them, for in this case they would not be able to perceive, observe and
describe meaningful events any longer—confronted with chaotic, meaningless

© 2005 FQS http://www.qualitative-research.net/fqs/

FQS 6(2), Art. 27, Udo Kelle: "Emergence" vs. "Forcing" of Empirical Data?
A Crucial Problem of "Grounded Theory" Reconsidered

and fragmented phenomena they would have to give up their scientific

endeavour. [5]

The infeasibility of an inductivist research strategy which demands an empty head

(instead of an "open mind") cannot only be shown by epistemological arguments,
it can also be seen in research practice. Especially novices in qualitative research
with the strong desire to adhere to what they see as a basic principle and
hallmark of Grounded Theory—the "emergence" of categories from the data—
often experience a certain difficulty: in open coding the search for adequate
coding categories can become extremely tedious and a subject of sometimes
numerous and endless team sessions, especially if one hesitates to explicitly
introduce theoretical knowledge. The declared purpose to let codes emerge from
the data then leads to an enduring proliferation of the number of coding
categories which makes the whole process insurmountable. In a methodological
self-reflection a group of junior researchers who had asked me for
methodological advice described this proliferation of code categories as follows:

"Especially the application of an open coding strategy recommended by Glaser and

Strauss—the text is read line by line and coded ad hoc—proved to be unexpectedly
awkward and time consuming. That was related to the fact that we were doing our
utmost to pay attention to the respondents' perspectives. In any case we wanted to
avoid the overlooking of important aspects which may lay behind apparently irrelevant
information. Our attempts to analyze the data were governed by the idea that we
should address the text tabula rasa and by the fear to structure data to much on the
basis of our previous knowledge. Consequently every word in the data was credited
with high significance. These uncertainties were not eased by advice from the
corresponding literature that open coding means a 'preliminary breaking down of
data' and that the emerging concepts will prove their usefulness in the ongoing
analysis. Furthermore, in the beginning we had the understanding that 'everything
counts' and 'everything is important'—every yet marginal incident and phenomenon
was coded, recorded in numerous memos and extensively discussed. This led to an
unsurmountable mass of data ..." (cf. KELLE et al. 2002, translation by UK). [6]

A more thorough look at the Discovery book reveals that GLASER and
STRAUSS were aware of that problem, since they wrote: "Of course, the
researcher does not approach reality as a tabula rasa. He must have a
perspective that will help him see relevant data and abstract significant categories
from his scrutiny of the data" (GLASER, STRAUSS 1967, p.3). [7]

GLASER and STRAUSS coined the term "theoretical sensitivity" to denote the
researcher's ability to "see relevant data", that means to reflect upon empirical
data material with the help of theoretical terms. "Sources of theoretical sensitivity
build up in the sociologist an armamentarium of categories and hypotheses on
substantive and formal levels. This theory that exists within a sociologist can be
used in generating his specific theory (...)" (ibid., p.46). But how can a researcher
acquire such an armamentarium of categories and hypotheses? The Discovery
book only contains a very short clue on the "great man theorists", which "(...)
have indeed given us models and guidelines for generating theory, so that with

© 2005 FQS http://www.qualitative-research.net/fqs/

FQS 6(2), Art. 27, Udo Kelle: "Emergence" vs. "Forcing" of Empirical Data?
A Crucial Problem of "Grounded Theory" Reconsidered

recent advances in data collection, conceptual systematization and analytic

procedures, many of us can follow in their paths" (p.11). One may find this re-
mark surprising given the sharp criticism of "theoretical capitalists" launched else-
where in the book. Furthermore the authors write that an empirically grounded
theory combines concepts and hypotheses which have emerged from the data
with "some existing ones that are clearly useful" (p.46). However, in the Discovery
book clear advice on how this combination can be pursued is missing. [8]

Consequently, in the most early version of Grounded Theory the advice to employ
theoretical sensitivity to identify theoretical relevant phenomena coexists with the
idea that theoretical concepts "emerge" from the data if researchers approach the
empirical field with no preconceived theories or hypotheses. Both ideas which
have conflicting implications are not integrated with each other in the Discovery
book. Furthermore, the concept of theoretical sensitivity is not converted into
clear cut methodological rules: it remains unclear how a theoretically sensitive
researcher can use previous theoretical knowledge to avoid drowning in the data.
If one takes into account the frequent warnings not to force theoretical concepts
on the data one gets the impression that a grounded theorist is advised to
introduce suitable theoretical concepts ad hoc drawing on implicit theoretical
knowledge but should abstain from approaching the empirical data with ex ante
formulated hypotheses. [9]

2. Different Approaches in Grounded Theory to Solve the Problem

2.1 GLASER's approach: theoretical coding with the help of "coding

families"

Much of GLASER's and STRAUSS' later methodological writings can be understood

as attempts to account for the "theoryladenness" of empirical observation and to
bridge the gap between "emergence" and "theoretical sensitivity". These attempts
followed two different lines: [10]

On the one hand, Barney GLASER tried to clarify the concept of "theoretical
sensitivity" in an own monograph published in 1978 with the help of the term
"theoretical coding", a process which he demarcates from "substantive coding".
Two different types of codes are linked to these different forms of coding:
substantive codes and theoretical codes. [11]

Substantive codes are developed ad hoc during "open coding", the first stage of
the coding process, and relate to the empirical substance of the research domain.
Theoretical codes which researchers always have to have at their disposal
"conceptualize how the substantive codes may relate to each other as
hypotheses to be integrated into a theory" (GLASER 1978, p.72). Theoretical
codes are used, in other words, to combine substantive codes to form a
theoretical model about the domain under scrutiny. The examples GLASER uses
for such theoretical codes are formal concepts from epistemology and sociology
which make basic claims about the ordering of the (social) world like the terms
causes, contexts, consequences and conditions: by calling certain events (which

© 2005 FQS http://www.qualitative-research.net/fqs/

FQS 6(2), Art. 27, Udo Kelle: "Emergence" vs. "Forcing" of Empirical Data?
A Crucial Problem of "Grounded Theory" Reconsidered

were coded with the help of substantive codes) as causes and others as
consequences or effects the hitherto developed substantive codes can be
integrated to a causal model. [12]

In the book "Theoretical Sensitivity" GLASER presents an extended list of terms

which can be used for the purpose of theoretical coding loosely structured in the
form of so called theoretical "coding families". Thereby various theoretical
concepts stemming from different (sociological, philosophical or everyday)
contexts are lumped together, as for example

• terms, which relate to the degree of an attribute or property ("degree family"),

like "limit", "range", "extent", "amount" etc.,
• terms, which refer to the relation between a whole and its elements
("dimension family"), like "element", "part", "facet", "slice", "sector", "aspect",
"segment" etc.,
• terms, which refer to cultural phenomena ("cultural family") like "social
norms", "social values", "social beliefs" etc.,

and 14 further coding families which contain terms from highly diverse theoretical
backgrounds, debates and schools of philosophy or the social sciences. Thereby
many terms can be subsumed under different "coding families": the term goal, for
instance, is part of a coding family referring to action strategies ("strategies
family") and also belongs to a coding family referring to the relation between
means and ends ("means-goal family"). [13]

Thus GLASER offers an equipment (one dares to say: a hotchpotch) of concepts

which are meant to guide the researcher in developing theoretical sensitivity but
fails to explain how such terms can be used and combined to describe and
explain empirical phenomena. That this task remains extremely difficult and can
hardly be achieved by applying single coding families can be easily shown with
regard to the first and most important coding family referring to causal relations.
The problem with that coding family is that general notions of cause and effect
can never sufficiently specify which types of events in a certain domain have to
be regarded as causes and which ones are to be seen as effects. Having terms
denoting causal relations (like "cause", "condition", "consequence" etc.) at hand
is in itself not sufficient for the development of causal models. Using such a
coding family one could consider in principle all events as causes and effects
which covary to a certain degree. To formulate a causal model about the relation
between certain specific events it would be necessary to use at least one
substantial (i.e. sociological, psychological ...) category in the development of a
causal explanation which provides a clue about which types of events regularly
covary. In order to develop theoretical models about empirical phenomena formal
or logical concepts (like "causality") have to be combined with substantial
sociological concepts (like "social roles", "identity", "culture"). A major problem
with GLASER's list of coding families is that it completely lacks such a
differentiation between formal and substantial notions. Thus the concept of
theoretical coding offers an approach to overcome the inductivism of early

© 2005 FQS http://www.qualitative-research.net/fqs/

FQS 6(2), Art. 27, Udo Kelle: "Emergence" vs. "Forcing" of Empirical Data?
A Crucial Problem of "Grounded Theory" Reconsidered

Grounded Theory, but its utility for research practice is limited, since it does not
clarify, how formal and substantial concepts can be meaningfully linked to each
other in order to develop empirically grounded theoretical models. [14]

2.2 STRAUSS' and CORBIN's approach: axial coding and the coding
paradigm

In his book "Qualitative Analysis for Social Scientists", published in 1987, Anselm
STRAUSS describes a more straightforward and less complicated way how
researchers may code empirical data with a theoretical perspective in mind. As
with earlier versions of Grounded Theory the analyst starts with open coding
"scrutinizing the fieldnote, interview, or other document very closely; line by line,
or even word by word. The aim is to produce concepts that seem to fit the data"
(STRAUSS 1987, p.28). Thereby STRAUSS notes certain difficulties novices
"have in generating genuine categories. The common tendency is simply to take
a bit of the data (a phrase or sentence or paragraph) and translate that into a
precis of it" (p.29). Such difficulties can be overcome by using the so called
"coding paradigm" "especially helpful to beginning analysts" (p.27). It consists of
four items, namely "conditions", "interaction among the actors", "strategies and
tactics" and "consequences", which can be used explicitly or implicitly to structure
the data and to clarify relations between codes. This coding paradigm can be
especially helpful during "axial coding" which "consists of intense analysis done
around one category at time in terms of the paradigm items" (p.32). [15]

This idea is developed further in "Basics of Qualitative Research", a book written

by Anselm STRAUSS and Juliet CORBIN in 1990. Like GLASER, STRAUSS and
CORBIN take into account the fact that any empirical investigation needs an
explicit or implicit theoretical framework which helps to identify categories in the
data and to relate them in meaningful ways. While GLASER had used a list of
more or less related sociological and formal terms for that purpose, STRAUSS
and CORBIN drew on one general model of action rooted in pragmatist and
interactionist social theory (cf. CORBIN 1991, p.36; STRAUSS 1990, p.7) to build
a skeleton or "axis" for developing grounded theories. This "paradigm model" is
used "to think systematically about data and to relate them in very complex ways"
(STRAUSS & CORBIN 1990, p.99) and for determining the main purpose of
theory construction: analysing and modelling action and interaction strategies of
the actors. Thereby, special emphasis is laid on the intentions and goals of the
actors and on the process character of human action and interaction. [16]

Drawing on GLASER's terminology one would regard STRAUSS' and CORBIN's

coding paradigm as an elaborated coding family which guides a certain
theoretical coding process (called "axial coding" by STRAUSS and CORBIN):
categories and concepts, developed during open coding are investigated whether
they relate to (1.) phenomena at which the action and interaction in the domain
under study are directed, (2.) causal conditions which lead to the occurrence of
these phenomena, (3.) attributes of the context of the investigated phenomena,
(4.) additional intervening conditions by which the investigated phenomena are
influenced, (5.) action and interactional strategies the actors use to handle the

© 2005 FQS http://www.qualitative-research.net/fqs/

FQS 6(2), Art. 27, Udo Kelle: "Emergence" vs. "Forcing" of Empirical Data?
A Crucial Problem of "Grounded Theory" Reconsidered

phenomena and (6.) the consequences of their actions and interactions. During
axial coding the analyst tries to find out which types of phenomena, contexts,
causal and intervening conditions and consequences are relevant for the domain
under study. If, for instance, social aspects of chronic pain are investigated the
researcher may try to identify typical action contexts which are relevant for
patients with chronic pain as well as characteristic patterns of pain management
strategies. Thereafter it can be examined which pain management strategies are
used by persons with chronic pain under certain conditions and in varying action
contexts. This may lead to the construction of models of action which capture the
variance of the observed actions in the domain under study and which can
provide the basis for a theory about action strategies generally pursued in certain
situations. [17]

Within this new and refined framework of Grounded Theory methodology,

STRAUSS and CORBIN also take a more liberal position concerning the role of
literature in the research process, maintaining that "all kinds of literature can be
used before a research study is begun ..." (STRAUSS & CORBIN 1990, p.56). [18]

3. The Split Between GLASER and STRAUSS in the 1990s

After having finished their cooperation in joint research projects GLASER and
STRAUSS followed different paths in their attempts to elaborate and clarify
crucial methodological tenets of Grounded Theory. Thus their approaches vary to
a considerable extent. In the year 1992 GLASER turned against STRAUSS' and
CORBIN's version of Grounded Theory in a monograph titled "Emergence vs.
Forcing: Basics of Grounded Theory Analysis", published in his private publishing
venture and written in an exceptionally polemic style. In this book he accuses
STRAUSS and CORBIN for having betrayed the common cause of Grounded
theory. The charge which is restated in various forms in this book and which
represents the crucial thread of GLASER's critique is that by using concepts such
as "axial coding" and "coding paradigms" researchers would "force" categories on
the data instead of allowing the categories to "emerge". Contrary to STRAUSS
and CORBIN, GLASER maintains that researchers following the "true path" of
Grounded Theory methodology have to approach their field without any precise
research questions or research problems ("He moves in with the abstract
wonderment of what is going on that is an issue and how it is handled", GLASER
1992, p.22) and insists that "there is a need not to review any of the literature in
the substantive area under study" (ibid., p.31). Following GLASER, the
application of theoretical background knowledge about the substantive field has
to be considered as harmful when developing grounded theories: "This dictum is
brought about by the concern to not contaminate, be constrained by, inhibit, stifle
or otherwise impede the researcher's effort to generate categories, their
properties, and theoretical codes" (ibid.). [19]

GLASER strictly affirms the inductivist rhetoric already put forward in the
Discovery book claiming that theoretical insights about the domain under scrutiny
would "emerge" directly from the data if and only if researchers free themselves
from any previous theoretical knowledge. However, GLASER's version of

© 2005 FQS http://www.qualitative-research.net/fqs/

FQS 6(2), Art. 27, Udo Kelle: "Emergence" vs. "Forcing" of Empirical Data?
A Crucial Problem of "Grounded Theory" Reconsidered

Grounded Theory takes into account basic problems of inductivism to a certain

extent: a strategy of scientific investigation which approaches an empirical
domain without any theoretical preconceptions is simply not feasible—such a
method would yield a plethora of incoherent observations and descriptions rather
than empirically grounded categories or hypotheses. The concepts of theoretical
sensitivity and theoretical codes can be seen as attempts to solve this
foundational epistemological problem. But theoretical sensitivity, the ability to
grasp empirical phenomena in theoretical terms, requires an extended training in
sociological theory (cf. GLASER 1992, p.28). Consequently, the "coding families"
proposed by GLASER in the book about theoretical sensitivity published in 1978
are of limited help for novices in empirical research who will have serious
difficulties to handle the more or less unsystematic list of theoretical terms from
various sociological and epistemological backgrounds offered by GLASER. And a
researcher with a broad and extended theoretical background knowledge and a
longstanding experience in the application of theoretical terms, on the other hand,
would certainly not need such a list. [20]

STRAUSS' and CORBIN's concept of a "coding paradigm" serves to explicate the

construction of theoretical framework necessary for the development of
empirically grounded categories in a much more user-friendly way. By drawing on
this concept researchers with limited experience in the application of theoretical
knowledge can use Grounded Theory methodology without taking the risk of
drowning in the data. One has to be very clear about the fact, however, that the
coding paradigm stems from a certain theoretical tradition, which is pragmatist
social theory rooted in the works of DEWEY and MEAD. Therefore GLASER's
suspicion that an application of the coding paradigm may lead to the "forcing" of
categories on the data cannot simply be dismissed. However, if one looks more
thoroughly at the conceptual design of STRAUSS' and CORBIN's coding
paradigm GLASER's critique seems to be overdrawn: the general theory of action
underlying the coding paradigm carries a broad and general understanding of
action which is compatible with a wide variety of sociological theories (ranging
e.g. from Rational Choice Theory to functionalist role theory or even sociological
phenomenology). It can be also argued that the "coding paradigm" to a great
extent represents an everyday understanding of purposeful and intentional
human action useful for the description of a wide array of social phenomena.
However, it must be noted here, that STRAUSS' and CORBIN's coding paradigm
is linked to a perspective on social phenomena prevalent in micro-sociological
approaches emphasizing the role of human action in social life. Researchers with
a strong background in macro-sociology and system theory may feel that this
approach goes contrary to their requirements and would be well advised to
construct an own coding paradigm rooted in their own theoretical tradition. [21]

GLASER's approach of "theoretical coding" whereby researchers introduce ad

hoc theoretical codes and coding families which they find suitable for the data
under scrutiny provides a strategy applicable for a greater variety of theoretical
perspectives. However, as has been said before following this strategy is much
more challenging esp. for novices since it lacks a readymade conceptual
framework like STRAUSS' and CORBIN's coding paradigm. However, it is

© 2005 FQS http://www.qualitative-research.net/fqs/

FQS 6(2), Art. 27, Udo Kelle: "Emergence" vs. "Forcing" of Empirical Data?
A Crucial Problem of "Grounded Theory" Reconsidered

interesting to note that GLASER's work obviously does not suggest a highly
pluralistic use of coding families (which would include the use of concepts from
macro-sociological approaches) since he seems to share STRAUSS' strong
inclination towards action and action theory; at least in his monograph
"Theoretical Sensitivity" he asserts that coding and coded incidents have to be
related to actions of the actors in the empirical domain. [22]

One of the most crucial differences between GLASER's and STRAUSS'

approaches of Grounded Theory lies in the fact that STRAUSS and CORBIN
propose the utilization of a specified theoretical framework based on a certain
understanding of human action, whereas GLASER emphasises that coding as a
process of combining "the analyst's scholarly knowledge and his research
knowledge of the substantive field" (GLASER 1978, p.70) has to be realised ad
hoc, which means that it has often to be conducted on the basis of a more or less
implicit theoretical background knowledge. Compared to this major dissimilarity,
other differences between the two approaches play a minor role. However,
GLASER seems to overstate some of them for rhetorical reasons. By highlighting
the "emergence" of theoretical concepts from the data he is drawn to
exaggerated truth claims: following GLASER the task of empirical research is the
discovery of social worlds and "facts" "as they really are". "In grounded theory (...)
when the analyst sorts by theoretical codes everything fits, as the world is
socially integrated and grounded theory simply catches this integration through
emergence" (GLASER 1992, p.84). Following such claims any attempt of further
examination of the "emerged" verities becomes superfluous and a falsification of
theoretical statements developed from the data simply would be impossible. This
would not only denounce the well established idea (which is now common
wisdom in almost any empirical science) that the purpose of empirical research is
not to discover unchangeable verities but to tentatively suggest and further cor-
roborate hypotheses, but also the epistemological insight that any empirical
phenomenon can be described in various ways and that any object can be
described and analysed under different theoretical perspectives. Instead it is
suggested that if and only if the analyst frees himself/herself from any theoretical
previous knowledge the "emerging" of categories from the data would ensure that
only relevant aspects of the phenomena under scrutiny are recognised and
described. This in fact represents a dogmatic inductivism prominent in early
empiricist philosophy—the conviction put forward for instance by BACON that
researchers having cleansed themselves from any theoretical preconceptions and
wrong "idols" and thus transformed the mind into a tabula rasa would gain the
ability to grasp empirical facts "as they really are". However, GLASER had made
clear elsewhere that theoretical concepts do not simply arise from the data alone
but through careful "theoretical coding" (that means: by categorizing empirical
data on the basis of previous theoretical knowledge). Thus the suspicion arises
that the parlance of "emergence" fulfils the function to legitimise a specific style of
research: under this perspective the "emergence talk" would not serve the
purpose to describe a methodological strategy but would simply offer a way to
immunise theories with the help of a methodological rhetoric: following this
rhetoric a researcher who follows the "right path" of Grounded Theory cannot go
wrong since the concepts have been emerged from the data. [23]

© 2005 FQS http://www.qualitative-research.net/fqs/

FQS 6(2), Art. 27, Udo Kelle: "Emergence" vs. "Forcing" of Empirical Data?
A Crucial Problem of "Grounded Theory" Reconsidered

4. Towards a Clearer Understanding of the "Grounding" of Categories

and Theories

From its beginnings the methodology of Grounded Theory has suffered from an
"inductivist self misunderstanding" entailed by some parts of the Discovery book.
Although this inductivism plays a limited role in research practice of many
Grounded Theory studies (including those of the founding fathers) it has often
lead to confusion especially among novices who draw their basic methodological
knowledge from text books. In the past decades Grounded Theory has made
considerable progress in overcoming the naïve empiricism of the emergence talk.
Thereby the concepts of "theoretical sensitivity", "theoretical coding", "axial
coding" and "coding paradigms" represent important steps in the development of
an adequate understanding of how qualitative data can be used in the process of
developing theoretical categories. Thus one can use Grounded Theory
procedures without adhering to the basic "dogmas of empiricism" (QUINE 1951)
namely the idea that at a certain stage of the research process a kind of
observation and description of empirical phenomena must take place which is not
"contaminated" by theoretical notions. However, inductivism still plays a vital role
in the image of Grounded Theory for a wider audience as well as in interior
methodological discussions, as the previous examples have shown. This leads to
the fact that many epistemologically informed social scientists repudiate
Grounded Theory after having read writings which seem to reject the trite
epistemological fact that there can be no empirical observations "unimpregnated
by expectations". [24]

In the following it will be shown that the explicit use and discussion of some
concepts nowadays widely discussed and well known in contemporary
methodology and epistemology could lead to a better understanding of the nature
of empirically grounded theory construction, especially since an implicit use of
these concepts already takes place and plays a role in Grounded Theory
methodology:

1. the concept of abductive (or retroductive) inference,

2. the concept of empirical content or falsifiability,
3. the concept of corroboration. [25]

4.1 Abductive inference as a logical foundation of theory building

In conceptualising the process of theory generation in empirical research often a

wrong alternative is established between an inductivist concept and a
hypothetico-deductive (H-D) model of theory generation: according to the H-D
model, favoured often by quantitative methodologists, research is seen as a
process of hypothesis testing by means of experimental or quasi-experimental
strategies. Following this view hypotheses cannot be derived from data, but
emerge from the researcher's speculations or happy guesses. The next step of
the research process would be rational elaboration of such hypotheses and the
operationalisation of their main elements, so that the hypotheses can be tested.

© 2005 FQS http://www.qualitative-research.net/fqs/

FQS 6(2), Art. 27, Udo Kelle: "Emergence" vs. "Forcing" of Empirical Data?
A Crucial Problem of "Grounded Theory" Reconsidered

Therefore, in the context of the H-D model the researcher has always to develop
precise hypotheses before collecting empirical data. Consequently, qualitative
research that implies the utilisation of unstructured data and the generation of
theories from that material would not be considered a rigorous and valid research
strategy from the viewpoint of the H-D model. [26]

However, since the 1970s a number of empirical investigations into the history of
science have shown that the H-D model cannot provide an adequate account of
the process of numerous scientific discoveries even in the Natural Sciences. As a
consequence, a lively discussion about the role of logics of discovery and rational
heuristics which has taken place in the modern philosophy of science has
challenged the view put forward by proponents of the H-D model that hypotheses
emerge through a process which is governed by mere speculation or "happy
guesses". Investigations into the history of natural sciences demonstrate that
scientific discoveries were in fact not only momentary mental episodes that are
not reconstructible as reasoning (cf. HANSON 1965; CURD 1980; NICKLES
1980, 1985, 1990). Although the context of discovery always contains elements
of intuition and creativity, the generation of a hypothesis can be reconstructed as
a reasoned and rational affair. In one of the most illuminating reconstructions of
scientific discoveries Norwood HANSON (1965) utilizes KEPLER's discovery of
the planetary orbits to show that logical inferences which lead to the discovery of
new theoretical insights are neither inductive nor deductive. Instead they
represent a special kind of logical reasoning whose premises are a set of
empirical phenomena and whose conclusion is an explanatory hypothesis. [27]

HANSON called this form of reasoning retroductive inference, in more recent

writings it has been also called "inference to the best explanation" (ACHINSTEIN
1992). One could also use the term "hypothetical reasoning" which reflects its
specific role in the research process: hypothetical inferences serve to discover a
hypothesis which explains certain empirical findings. [28]

The earliest concepts of hypothetical reasoning were developed by the

pragmatist philosopher Charles Sanders PEIRCE who described a third form of
inference apart from deduction and induction which he called "hypothesis" or
"abduction". Deductive reasoning is the application of general rules to specific
cases to infer a result.

"The so-called major premise lays down this rule; as for example, 'All men are
mortal'. The other or minor premise states a case under the rule; as 'Enoch was a
man'. The conclusion applies the rule to the case and states the result: 'Enoch is
mortal'" (PEIRCE 1974/1979, 2.621). [29]

Induction is an inversion of this deductive syllogism—by induction one generalises

from a number of cases where a certain result is observed, and infers to general
rule, claiming that these results can be observed in all cases of a class which the
observed cases belong to. Another way of inverting a deductive syllogism is
hypothetical inference which starts with an empirical phenomenon and proceeds
to a general statement which explains the observed phenomenon. Thereby the

© 2005 FQS http://www.qualitative-research.net/fqs/

FQS 6(2), Art. 27, Udo Kelle: "Emergence" vs. "Forcing" of Empirical Data?
A Crucial Problem of "Grounded Theory" Reconsidered

researcher either has a general rule at his disposal that leads to a possible
explanation, or the hypothetical inference serves as a means to discover new,
hitherto unknown concepts or rules. Often such an "abductive" inference (cf.
REICHERTZ 2003) starts by a surprising, anomalous event which cannot be
explained on the basis of previous knowledge: "The surprising fact, C is
observed. But if A were true, C would be a matter of course. Hence there is a
reason to suspect that A is true" (PEIRCE 1974/1979, 5.189). [30]

Confronted with an anomalous event "we turn over our recollection of observed
facts; we endeavour so to rearrange them, to view them in such new perspective
that the unexpected experience shall no longer appear surprising" (PEIRCE
1974/1979, 7.36). This is, of course, a creative endeavour which sometimes
"comes to us like a flash" (PEIRCE 1974/1979, 5.182). Nevertheless, the
researcher's creativity is limited by certain constraints and methodological rules.
First of all, the originality of the newly developed hypotheses is limited by the
facts which must be explained. "It is not pure, ontological originality in the relation
to the ideas and perceptual facts at hand. Hypotheses can be original, but only if
they still may explain the facts in question" (ANDERSON 1987, p.44).
Furthermore, an abductive inference must not only lead to a satisfactory
explanation of the observed facts but must be related to the previous knowledge
of the researcher—"the different elements of the hypothesis were in our minds
before", as PEIRCE put it (1903, 5.181). For that reason abductions do not lead
to the creation of new knowledge ex nihilo. Instead, every new insight combines
"something old and something hitherto unknown" (7.536). Abduction becomes an
innovative process by modifying and combining several elements of previous
knowledge—"it is the idea of putting together what we had never before dreamed
of putting together which flashes the new suggestion before our contemplation"
(5.182). Scientific discoveries always require the integration of previous
knowledge and new experience "(...) that is to say, we put old ideas together in a
new way and this reorganization itself constitutes a new idea" (ANDERSON
1987, p.47). Many of the theoretical insights and developments in sociology which
led to new and convincing explanations of social phenomena may be
reconstructed as arising from abductive inferences. This esp. relates to so called
"middle range theories", as for instance DURKHEIM's idea that differences
between suicide rates result from differing levels of "anomia", or WEBER's
explanation of the economic success of protestant merchants as a consequence
of their religious orientations. The "labelling approach" which attempted to
understand "mental illness" or deviance not as an inherent personal quality or
attribute of individual actors but as a result of processes of social interaction may
serve as another good example. All these theoretical explanations of social
phenomena which mark significant theoretical advancements in sociology started
with sometimes surprising, anomalous or difficult empirical phenomena which
were explained by drawing on theoretical concepts or ideas previously not applied
to the domain under scrutiny: thus WEBER related success in worldly affairs to
religious beliefs referring to transcendent realities. Or the proponents of the
labelling approach interpreted odd or problematic behaviour as a result of
interactive processes of role definition and identity formation. In making abductive
inferences, researchers depend on previous knowledge that provide them with

© 2005 FQS http://www.qualitative-research.net/fqs/

FQS 6(2), Art. 27, Udo Kelle: "Emergence" vs. "Forcing" of Empirical Data?
A Crucial Problem of "Grounded Theory" Reconsidered

the necessary categorical framework for the interpretation, description and

explanation of the empirical world under study. If an innovative research process
should be successful this framework must not work as a Procrustean bed into
which empirical facts are forced. Instead, the framework which guides empirical
investigations should be modified, rebuilt and reshaped on the basis of empirical
material. [31]

4.2 Empirical content or falsifiability as a criterion for the applicability of

theoretical preconceptions in qualitative inquiry

Hypothetical inferences combine new and interesting empirical facts with existing
theoretical knowledge in a creative way. By no means that does imply that the
theoretical knowledge of the qualitative researcher should form in the beginning a
fully coherent network of explicit propositions from which precisely formulated and
empirically testable statements can be deduced. Rather it should constitute (a
sometimes only loosely connected) "heuristic framework" of concepts (or "coding
families") which helps the researcher to focus the attention on certain phenomena
in the empirical field. But doesn't that mean that theoretical sensible categorising
and "coding" of data is merely a gift of charismatic researchers? Can certain
aspects of it be made explicit, for instance by determining relevant "theoretical
codes" before the data are coded? Is the construction and use of an (at least
partly) predefined category scheme a sensible strategy in qualitative analysis or
does this necessarily seduce the researcher to go astray so that he/she
abandons basic principles of qualitative research, namely the principles of
discovering new patterns and relations? [32]

To solve this problem it is helpful to discuss a concept which plays an important

role in the writings of Karl POPPER and other traditional proponents of the H-D
model: "falsifiability" or "empirical content". This concept is normally used to
identify sound scientific hypotheses in a H-D framework. In this context one
regards only clear-cut and precisely formulated propositions with empirical
content as adequate hypotheses whereas concepts and hypotheses which lack
empirical content and thus cannot be falsified are considered as highly
problematic. Theoretical concepts with low empirical content, however, can play
an extremely useful role if the goal of empirical research is not the testing of
predefined hypotheses but the empirically grounded generation of theories, since
they do not force data into a Procrustean bed—their lack of empirical content
gives them flexibility so that a variety of empirical phenomena can be described
with their help. Although such concepts cannot be "tested" empirically, they may
be used as heuristic concepts which represent "lenses" through which researcher
perceive facts and phenomena in their research field. [33]

Two different types of such heuristic concepts may be used to define a category
scheme useable for the structuration and analysis of qualitative data which can
be supplemented, refined and modified in the ongoing process of empirical
analysis: [34]

© 2005 FQS http://www.qualitative-research.net/fqs/

FQS 6(2), Art. 27, Udo Kelle: "Emergence" vs. "Forcing" of Empirical Data?
A Crucial Problem of "Grounded Theory" Reconsidered

The first important type of heuristic concept refers to a variety of theoretical

notions, definitions and categories drawn from "grand theories" in the social
sciences which are too broad and abstract to directly deduce empirically
contentful propositions. Herbert BLUMER invented the term "sensitizing
concepts" to describe theoretical terms which "lack precise reference and have
no bench marks which allow a clean cut identification of a specific instance"
(BLUMER 1954, p.7). Sensitizing concepts are useful tools for descriptions but
not for predictions, since their lack of empirical content permits researchers to
apply them to a wide array of phenomena. Regardless how empirically
contentless and vague they are, they may serve as heuristic tools for the
construction of empirically grounded theories. [35]

A concept like "role-expectations" can serve as a good example for that. The
assertion that individuals act in accordance with role expectations does not imply
a lot of information by itself. This concept may, however, be useful to formulate a
variety of research questions for the investigation of different substantive fields:
Do role expectations play an important role in the empirical domain under study?
What kind of role expectations can be found? By which means do empirical
actors try to meet them? Do certain actors develop strategies to avoid the fulfil-
ment of role expectancies? Are such strategies revealed by other actors in the
investigated field? Etc. Concepts from so called "utility theory" may serve as
another example: at the core of utility theory is the idea that human actors will
choose the action which seems the most adequate for the achievement of a
desired goal from a set of given action alternatives. However, without specifying
which goals the actors pursue and which actions they consider to be adequate,
such a proposition has no empirical content. The theory is like an "empty sack"
(cf. SIMON 1985), if one does not specify further auxiliary assumptions. Instead
of allowing for the development of precise hypotheses utility theory may provide
researchers with useful research questions and heuristic codes: qualitative
researchers may, for instance, code text segments which refer to the potential
costs and benefits that certain actions may have for the actors, they may code
segments which relate to the intentions and goals of the research subjects or the
means they use to reach their goals etc. In this manner researchers can draw on
a wide variety of abstract notions from different theoretical traditions to structure
their data. But one should never forget in this process that sticking to certain
theoretical tradition makes it easier to structure the data but also carries the risk
that concepts are neglected that would suit the data even better and would yield
more interesting insights. Even sensitizing and heuristic concepts that capture all
kinds of different phenomena may lead to an exclusion of other theoretical
perspectives: thus the extended use of concepts with a strong background in
micro-sociological action theory (e.g. "actor", "purposes" ...) can preclude a
system theory and macro-perspective. [36]

A strategy to cope with that risk (better suited than the waiting for an
"emergence" of the most adequate theoretical notions from the data) would be
the use of different and even competing theoretical perspectives on the same
data. Furthermore, special attention should be paid to the question whether a

© 2005 FQS http://www.qualitative-research.net/fqs/

FQS 6(2), Art. 27, Udo Kelle: "Emergence" vs. "Forcing" of Empirical Data?
A Crucial Problem of "Grounded Theory" Reconsidered

chosen theoretical concept leads to the exclusion or neglect of certain

phenomena and incidents contained in the data. [37]

A second type of categories which do not easily force data but allow for the
discovery of previously unknown relations and patterns are categories which
relate to general topics of interest covered in the data material. Such topic
oriented categories can be often easily found by drawing on general common
sense knowledge or on specific local knowledge of the investigated field.
Categories like "school", "work" or "family" represent simple examples for that,
but topic oriented categories may be far more complex. However, one should
always ask the question, as with heuristic theoretical concepts, whether a certain
code really serves for heuristic purposes or whether it excludes relevant
phenomena from examination. [38]

Both types of heuristic categories, categories developed from common sense

knowledge as well as categories derived from abstract theoretical concepts fit
various kinds of social reality. That means: it is not necessary to know concrete
facts about the investigated domain in order to start using these concepts for data
analysis. In other words: heuristic categories cannot be used to construct
empirically contentful propositions without additional information about empirical
phenomena. This makes them rather useless in the context of an H-D strategy,
but it is their strength in the context of exploratory, interpretative research.
Regardless whether heuristic categories are derived from common-sense
knowledge or from abstract theoretical concepts the following rule is always
applicable: with decreasing empirical content the risk that the data are "forced" is
diminished. [39]

Thus the epistemological concept of "empirical content" and "falsifiability" can

help to identify preconceptions which qualitative researchers (whether they apply
Grounded Theory methodology or not) may use to structure the data material
while minimising the risk to violate basic methodological concepts of qualitative
research. Previous theoretical knowledge can be used at any stage of the
process of empirically grounded theory construction if the researchers draw on
theoretical concepts with limited empirical content (which the H-D approach would
dissuade us to use). Thereby, the researcher may start qualitative analysis by
using heuristic concepts and may then proceed to the construction of categories
and propositions with growing empirical content. In this process grand theories
play the role of a theoretical axis or a "skeleton" to which the "flesh" of empirically
contentful information from the research field is added in order to develop
empirically grounded categories and propositions. [40]

However, in some cases also the use of categories and assertions with high
empirical content can prove to be fruitful in a qualitative study. A researcher
investigating the process of care-giving to frail and elderly people, for instance,
may discover that Arlie HOCHSCHILD's concept of "emotional labour" (1983)
turns out to be helpful in the understanding of phenomena in the research
domain. This concept was initially developed to describe typical patterns of action
and interactions of flight attendants and air passengers but can be transferred to

© 2005 FQS http://www.qualitative-research.net/fqs/

FQS 6(2), Art. 27, Udo Kelle: "Emergence" vs. "Forcing" of Empirical Data?
A Crucial Problem of "Grounded Theory" Reconsidered

other domains of social services. Obviously this concept comprises more

empirical content than the term "role expectation"—compared to the latter term
"emotional labour" cannot be related to any social interaction. There are obviously
social interactions which do not require emotional labour, and the assertion that
certain service providers are expected to do emotional labour can in principle be
falsified. On the other hand, the concept can be rather illuminating in
understanding social relations in various fields. [41]

Consequently, it can be sensible in qualitative research to sometimes also use

concepts which are closer to the understanding of the term "theory" in H-D
research: definite categories and propositions about a certain field of social action
that entail enough empirical content to be tested. There is no reason to abstain
from such concepts, esp. since their use represents a long and well-established
tradition in qualitative research. Researchers and methodologists coming from
the "Chicago School" of American sociology had proposed in the 1930s a
research strategy named "Analytic Induction" which was used thereafter in many
famous qualitative studies. Thereby initial hypotheses are examined and modified
with the help of empirical evidence provided by so called "crucial cases". A well-
known example comes from Donald CRESSEY's qualitative study about em-
bezzlement. During his research, for instance, he formulated the hypothesis that

"... trust violators usually consider the conditions under which they violated their own
positions of trust as the only "justifiable" conditions, just as they consider their own
trust violation to be more justified than a crime such as robbery or burglary"
(CRESSEY 1973, pp.104f.)

—a statement which can in principle be falsified, if one undertakes the effort of

collecting data about trust violators. At a certain point in the research process
CRESSEY indeed searched systematically for "crucial cases" and "negative
instances" of trust violators who saw their trust violations as justifiable. [42]

However, by applying such a research strategy there is always the risk that data
are structured with the help of concepts which are not suited for the specific
research domain and which do not match the researcher's theoretical interests
and orientations. The already mentioned risk that the heuristic concepts
employed may contain too much empirical content for the researcher's purposes
is already prevalent with STRAUSS' coding paradigm which can draw qualitative
researchers towards a certain micro-sociological orientation which they do not
necessarily share. On the other hand, the advice to use categories with low
empirical content may be unhelpful for inexperienced researchers, since in a
given research domain not every heuristic concept can draw the researcher's
attention to sociologically relevant phenomena and thus yield insights and
interesting results. This danger may arise with GLASER's "coding families": it can
be a highly demanding task esp. for novices to select the theoretical concept
most suited for a certain research domain among a choice of numerous
theoretical schools and approaches. [43]

© 2005 FQS http://www.qualitative-research.net/fqs/

FQS 6(2), Art. 27, Udo Kelle: "Emergence" vs. "Forcing" of Empirical Data?
A Crucial Problem of "Grounded Theory" Reconsidered

An important task of qualitative methodology would be to show a middle path

between the "Scylla" of forcing the data with preconceived notions and theories
not suited for the domain under study and the "Charybdis" of an indiscriminate
and eclectic use of concepts from various theoretical traditions. The following
methodological strategies can help researchers to avoid either danger:

• The development of empirically grounded categories and hypotheses benefits

from theoretical pluralism. A pluralistic use of heuristic frameworks requires
that researchers have a variety of different concepts with diverging theoretical
background at their disposal and obtain a flexible choice among them after
having examined their appropriateness for the investigated phenomena.
Experts with longstanding experience may be able to choose the right
heuristic concept intuitively thereby drawing on rich theoretical background
knowledge. In contrast to that novices may benefit from an explicit style of
theory building in which different "grand theories" are utilised in order to
understand, explain and describe phenomena under study. A systematic
comparison of the results from the use of different heuristic concepts is by all
means preferable to an "emergence talk" which masks the use of the
researcher's pet concepts.
• A strategy already suggested by proponents of "Analytic Induction", the
systematic search for counter evidence, can reveal whether a given heuristic
concept has high or low empirical content. If negative instances are easily
found the applied categories obviously have a high degree of falsifiability or
empirical content and may be not suited as heuristic concepts which are used
in an initial attempt to structure empirical data.
• The same holds true for an extensive search for empirical phenomena to
which the used categories do not apply. If a variety of phenomena can be
identified which cannot be covered by heuristic concepts used so far it is
obviously necessary to look for alternative concepts which are suited better to
capture the investigated phenomena. [44]

4.3 The necessity of corroboration of empirically grounded categories and

hypotheses

Contrary to an inductivist understanding a model of the research process based

on "hypothetical" or "abductive inference" is consistently fallibilistic, that means
that it does not claim that the validity of propositions developed on the basis of
empirical data can be simply ascertained by the fact that the researcher has freed
the mind from any preconceptions whatsoever before collecting these data.
Hypothetical inferences may lead to rational and well-founded assertions which
are both consistent with observed phenomena and with previous theoretical
knowledge. If these assertions are not only mere descriptions of observed events
but represent theoretical claims they have to be regarded as fallible. The fallibility
of any theoretical claim developed on the basis of empirical observation via
hypothetical inferences can easily be seen from the fact that often one empirical
phenomenon allows for several theoretical explanations which are contradictory
but equally compatible with existing stocks of knowledge. [45]

© 2005 FQS http://www.qualitative-research.net/fqs/

FQS 6(2), Art. 27, Udo Kelle: "Emergence" vs. "Forcing" of Empirical Data?
A Crucial Problem of "Grounded Theory" Reconsidered

If one abandons the idea that definite and absolute reliable knowledge can be
developed from empirical data via induction and if one explicitly acknowledges
the role of previous theoretical knowledge in the research process one must also
not consider the demand to further corroborate empirically grounded theoretical
concepts as an attempt to downplay or underestimate the role of exploratory
inquiry compared to methods of (experimental or quasi-experimental) hypothesis
testing. This requirement rather represents a matter of course given the
methodological fact that empirical research can never provide a final proof for
theoretical propositions but only cumulative and always provisional evidence.
Whereas STRAUSS and CORBIN pay a lot of attention to the question how
grounded categories and propositions can be further validated, GLASER's
concept shows at least a gleam of epistemological fundamentalism (or "certism",
LAKATOS 1978) especially in his defence of the inductivism of early Grounded
Theory. "Grounded theory looks for what is, not what might be, and therefore
needs no test" (GLASER 1992, p.67). Such sentences carry the outmoded idea
that empirical research can lead to final certainties and truths and that by using
an inductive method the researcher may gain the ability to conceive "facts as they
are" making any attempt of further corroboration futile. [46]

If one does not want to adventure on claiming infallibility for particular results of
empirical research the further examination, modification and rejection of
empirically grounded hypotheses become an important issue. One may not only
draw on STRAUSS' and CORBIN's more current writings about the methodology
of Grounded Theory for that but can also use many concepts developed
throughout the history of qualitative research, e.g. the already mentioned strategy
of "Analytic Induction", procedures for the examination of hypotheses in
hermeneutic text interpretation in which different "Lesarten" (reading versions) of
the same text passage are corroborated through sequential analysis of additional
text (OEVERMANN et al. 1979) or methods for developing and testing causal
hypotheses in qualitative research proposed by Charles RAGIN (1987).
Techniques developed in the past two decades for a computer-assisted
categorisation, archiving and structuration of qualitative data can also support the
process of further grounding theoretical concepts in the data by systematically
searching for empirical evidence and counter-evidence (KELLE 2004). [47]

5. Conclusive Remarks

"Emergence" has turned out to be a rather problematic methodological concept

which reflects the empiricist idea that researchers can and must approach data
with no theories in mind. However, GLASER and STRAUSS did not overlook the
fact that researchers always have to draw on their existing theoretical knowledge
in order to understand, describe and explain empirically observed social
phenomena. An alternative to an inductivist understanding of qualitative research
can already be found in the Discovery book: the researcher's "theoretical
sensitivity" provides a "perspective that will help (him) see relevant data and
abstract significant categories from his scrutiny of the data" (ibid., p.3). Thus the
earliest version of Grounded Theory contained two different concepts concerning
the relation between data and theory with conflicting implications: on the one

© 2005 FQS http://www.qualitative-research.net/fqs/

FQS 6(2), Art. 27, Udo Kelle: "Emergence" vs. "Forcing" of Empirical Data?
A Crucial Problem of "Grounded Theory" Reconsidered

hand the idea is stressed that theoretical concepts "emerge" from the data if the
researcher approaches the empirical field with no preconceived theories or
hypotheses, on the other hand the researcher is advised to use his or her
previous theoretical knowledge to identify theoretical relevant phenomena in the
data. [48]

Much of GLASER's and STRAUSS' later methodological work can be understood

as attempts to further develop the concept of theoretical sensitivity in order to
reconcile these prima facie divergent ideas. Thereby STRAUSS proposes the use
of a general theory of action to build an axis of the emerging theory. GLASER,
although he had fully repudiated STRAUSS' concepts in 1992, proposed a similar
idea in 1978: theoretical codes represent those theoretical concepts which the
researcher has at his or her disposal independently from data collection and data
analysis. Thus the controversy between GLASER and STRAUSS boils down to
the question whether the researcher uses a well defined "coding paradigm" and
always looks systematically for "causal conditions", "phenomena", "context",
"intervening conditions", "action strategies" and "consequences" in the data, or
whether he or she should employ theoretical codes ad hoc, thereby drawing on a
huge fund of "coding families". [49]

Both strategies have their pros and cons: novices who wish to get clear advice on
how to structure data material may be satisfied with the use of the coding para-
digm. Since the paradigm consists of theoretical terms which carry only limited
empirical content the risk is not very high that data are forced by its application.
However, it must not be forgotten that it is linked to a certain micro-sociological
perspective. Many researchers may concur with that approach esp. since
qualitative research always had a relation to micro-sociological action theory, but
others who want to employ macro-sociological and system theory perspective
may feel that the use of the coding paradigm would lead them astray. [50]

Experienced researchers with a broad knowledge in social theory would clearly

benefit from the advantages of theoretical coding—having at their disposal not
only one possible axis of the developing theory but being able to construct such
an axis by themselves through the combination of theoretical concepts from
different schools of thought. But regardless of which types of "theoretical codes"
or "coding paradigms" are applied empirically grounded theory building should
always be guided by an adequate epistemological understanding of the relation
between data and theory. Thereby it is of utmost importance to abandon
inductivist rhetoric and to develop a clear understanding of the role of inductive
and abductive inferences in the process of empirically grounded theory
generation. Furthermore the insight must be stressed that any scientific discovery
requires the integration of previous knowledge and new empirical observations
and that researchers always have to draw on previous theoretical knowledge
which provides categorical frameworks necessary for the interpretation,
description and explanation of the empirical world. [51]

To make sure that by applying theoretical knowledge one does not force data into
a Procrustean bed one needs to thoroughly differentiate between diverse types of

© 2005 FQS http://www.qualitative-research.net/fqs/

FQS 6(2), Art. 27, Udo Kelle: "Emergence" vs. "Forcing" of Empirical Data?
A Crucial Problem of "Grounded Theory" Reconsidered

theoretical statements (namely between definite and precise hypotheses on the

one hand and broad and general heuristic concepts on the other hand) and their
differing role in the process of theory generation. Empirically grounded theory
building starts by making a careful choice among a variety of concepts with
diverging theoretical backgrounds after having examined their appropriateness
for the investigated phenomena. By using such a heuristic framework as the axis
of the developing theory one carefully proceeds to the construction of categories
and propositions with growing empirical content. This should be accompanied by
a meticulous search for negative instances and for empirical phenomena to which
the used heuristic categories do not apply and which would call for their
reformulation or abandonment. This style of inquiry should be supplemented by
strategies of further corroboration of the empirically contentful categories and
propositions developed in the ongoing course of theory building. [52]

References

Achinstein, Peter (1992). Inference to the Best Explanation: Or, Who Won the Mill-Whewell
Debate? Studies in the History and Philosophy of Science, 23, 349-364.
Anderson, Douglas R. (1987). Creativity and the Philosophy of C.S.Peirce. Dordrecht: Martinus
Nijhoff.
Blumer, Herbert (1954). What is wrong with Social Theory? American Sociological Review, 19,
3-10.
Corbin, Juliet (1991). Anselm Strauss: An Intellectual Biography. In David R. Maines (Eds.), Social
Organization and Social Process. Essays in Honor of Anselm Strauss (pp.17-44). New York: Aldine
de Gruyter.
Cressey, Donald R. (1950). The Criminal Violation of Financial Trust. American Sociological
Review, 15, 738-743.
Cressey, Donald R. (1953/1971). Other People's Money. A Study in the Social Psychology of
Embezzlement. Belmont: Wadsworth.
Chalmers, Alan F. (1999). What is this thing called science? Maidenhead: Open University Press.
Curd, Martin V. (1980). The logic of discovery: an analysis of three approaches. In Thomas Nickles
(Ed.), Scientific Discovery, Logic and Rationality (Boston Studies in the Philosophy of Science, Vol.
LVI) (pp.201-219). Reidel: Dordrecht.
Glaser, Barney (1978). Theoretical Sensitivity. Advances in the Methodology of Grounded Theory.
Mill Valley, Ca.: The Sociology Press.
Glaser, Barney (1992). Emergence vs. Forcing: Basics of Grounded Theory Analysis. Mill Valley,
Ca.: Sociology Press.
Glaser, Barney & Strauss, Anselm (1967). The Discovery of Grounded Theory: Strategies for
Qualitative Research. New York: Aldine de Gruyter.
Hanson, Norwood Russell (1965). Patterns of Discovery. An Inquiry Into the Conceptual
Foundations of Science. Cambridge: Cambridge University Press.
Hochschild, Arlie (1983). The Managed Heart. Commercialization of Human Feeling. Berkeley:
University of California Press.
Kelle, Udo (1995). Theories as Heuristic Tools in Qualitative Research. In Ilja Maso, Paul A.
Atkinson, Sarah Delamont & Jef C. Verhoeven (Eds.), Openness in Research. The tension
between Self and Other (pp.33-50). Assen: Van Gorcum.
Kelle, Udo (2004). Computer Assisted Qualitative Data Analysis. In David Silverman, Giampetro
Gobo, Clive Seale & Jaber F. Gubrium (Eds.), Qualitative Research Practice (pp.473-489). London:
Sage.
Kelle, Udo; Marx, Janine; Pengel, Sandra; Uhlhorn, Kai & Witt, Ingmar (2002). Die Rolle
theoretischer Heuristiken im qualitativen Forschungsprozeß – ein Werkstattbericht. In Hans-Uwe

© 2005 FQS http://www.qualitative-research.net/fqs/

FQS 6(2), Art. 27, Udo Kelle: "Emergence" vs. "Forcing" of Empirical Data?
A Crucial Problem of "Grounded Theory" Reconsidered

Otto, Gertrud Oelerich & Heinz-Günter Micheel (Eds.), Empirische Forschung und Soziale Arbeit.
Ein Lehr- und Arbeitsbuch (pp.11-130). Neuwied und Kriftel: Luchterhand.
Lakatos, Imre (1978). The Methodology of Scientific Research Programmes. Cambridge:
Cambridge University Press.
Laudan, Larry (1977). Progress and its Problems. Towards a Theory of Scientific Growth. London
and Henley: Routledge & Kegan Paul.
Lindesmith, Alfred R. (1947/1968). Addiction and Opiates. Chicago: Aldine.
Nickles, Thomas (ed.) (1980). Scientific Discovery, Logic and Rationality (Boston Studies in the
Philosophy of Science, Vol. LVI). Reidel: Dordrecht.
Nickles, Thomas (1985). Beyond Divorce: Current status of the discovery debate. Philosophy of
Science, 52, 177-206.
Nickles, Thomas (1990). Discovery Logics. Philosophica, 45, 732.
Oevermann, Ulrich; Allert, Tilmann; Konau, Elisabeth & Krambeck, Jürgen (1979). Die
Methodologie einer "objektiven Hermeneutik" und ihre allgemeine forschungslogische Bedeutung in
den Sozialwissenschaften. In Hans-Georg Soeffner (Ed.), Interpretative Verfahren in den Sozial-
und Textwissenschaften (pp.352-434). Stuttgart: Metzler.
Peirce, Charles S. (1974, 1979). Collected Papers. Published by Charles Hartshore, Paul Weiss
and Arthur Burks. Cambridge (Mass.): The Belknap Press of Harvard University Press.
Quine, Willard Orman von (1951). Two Basic Dogmas of Empiricism. The Philosophical Review, 60,
20-43.
Ragin, Charles (1987). The Comparative Method. Moving beyond Qualitative and Quantitative
Methods. Berkeley: University of California Press.
Reichertz, Jo (2003). Die Abduktion in der qualitativen Sozialforschung. Opladen: Leske und
Budrich.
Simon, Herbert A. (1985). Human nature in politics: the dialogue of psychology with political
science. The American Political Science Review, 79, 293-304.
Strauss, Anselm L. (1970). Discovering New Theory from previous Theory. In Tamotsu Shibutani
(Ed.), Human Nature and Collective Behavior: Papers in Honor of Herbert Blumer (pp.46-53).
Englewood Cliffs: Prentice Hall.
Strauss, Anselm L. (1987). Qualitative Analysis for Social Scientists. Cambridge, NY.: Cambridge
University Press.
Strauss, Anselm L. (1990). Creating Sociological Awareness. New Brunswick: Transaction Publ.
Strauss, Anselm L. & Corbin, Julliet (1990). Basics of Qualitative Research. Grounded Theory
Procedures and Techniques. Newbury Park, Ca.: Sage.

Author

Udo KELLE has been recently appointed as Contact:

Professor for Social Research Methods at the
Department of Social Sciences and Philosophy at PD Dr. Udo Kelle
the University of Marburg. He has written various Institute for Sociology
books and articles about the methodology of Philipps-University of Marburg
qualitative research and about the integration of Ketzerbach 11
qualitative and quantitative methods (e.g. 35032 Marburg, Germany
"Computer-aided Qualitative Data Analysis",
London, Sage 1995 or (with Susann KLUGE) E-mail: udo.kelle@staff.uni-marburg.de
"Vom Einzelfall zum Typus", Opladen, Leske and
Budrich 1999, or "Methodeninnovation in der
Lebenslaufforschung", Opladen, Juventa 2001).
His main research interests cover the
methodological and philosophical background of
social research methods and the connections
between sociological theory and data in empirical
research. His current work involves extensive
research in the field of Mixed Methods Designs
and their application in the field of sociological life
course research.

© 2005 FQS http://www.qualitative-research.net/fqs/

FQS 6(2), Art. 27, Udo Kelle: "Emergence" vs. "Forcing" of Empirical Data?
A Crucial Problem of "Grounded Theory" Reconsidered

Citation

Kelle, Udo (2005). "Emergence" vs. "Forcing" of Empirical Data? A Crucial Problem of "Grounded
Theory" Reconsidered [52 paragraphs]. Forum Qualitative Sozialforschung / Forum: Qualitative
Social Research, 6(2), Art. 27, http://nbn-resolving.de/urn:nbn:de:0114-fqs0502275.
Revised 3/2007

© 2005 FQS http://www.qualitative-research.net/fqs/